Background. Malnutrition and infection commonly co-exist. Global efforts to improve child health have been guided by the understanding that this relationship is bi-directional and causal. Vast resources have been invested in the prevention and management of infectious diseases with the belief that reductions in malnutrition would naturally follow. In spite of global progress in controlling infection, continuing high levels of malnutrition have led to a re-thinking of this policy and a re- examination of the malnutrition/infection complex. Previous analyses have often used inappropriate statistical methods and failed to consider broader socioeconomic influences. Objectives and Aims. The overall objective of the proposed project is to understand the causal relationships among malnutrition, infection, and diet in pre-school children living in impoverished communities. The primary substantive aim is to determine the direction, and quantify the relative importance of these relationships. Several hypotheses will be tested towards this aim. For example, it is hypothesized that diet is an effect modifier of the impact of infectious disease on nutritional status: diarrhea has a negative impact on the nutritional status of children with poor diets while it has little impact on nutritional status of those with adequate diets. The primary methodological aim is to compare the results of standard epidemiologic approaches (e.g. OLS regression) with those of alternative approaches, particularly structural equation modelling (SEM) for testing these hypotheses. Methods. Dataset Data from a recently (1987-1991) conducted, community- based, prospective longitudinal study of about 500 Guatemalan children will be initially used to test the hypotheses. Detailed information on nutritional status, growth, illness, treatments, diet, and family characteristics is available making it one of the best longitudinal studies of child health and infection in existence. The principal Investigator (D.S.) has worked with this dataset in the past and has received formal permission to take full advantage of this resource. Analysis: The relationship between malnutrition and diarrhea is conceptualized as being bi-directional and variable over the life of the child. In addition, it is hypothesized that this relationship works within, and is modified by, a system of underlying conditions (e.g. maternal education) and proximate determinants (e.g. feeding practices). Structural equation modelling will therefore be used to simultaneously estimate the parameters associated with this multilevel, nonrecursive, multiwave system. In years 4 and 5 of the project, confirmation of these models will be achieved using data from a separate longitudinal study of the generational effects of malnutrition currently underway in Guatemala. Significance. Health: Approximately 14 million children die each year, most due to the synergistic effects of infection and malnutrition. A better understanding of the malnutrition/infection complex will allow policy makers to design more effective interventions towards improving global health and survival. Methodological: Although structural equation modelling (SEM) is frequently used in the social sciences, its use in the biomedical field is rare. Comparing SEM with standard epidemiological techniques will provide insight into the usefulness of SEM in other areas of health research (eg. HIV/AIDS) that involve multilevel determinants of reciprocal causal relationships.